[MUD-Dev] openmud: open issues

[James Wilson]

I'd like to set down the design issues that have come up, in short pithy blurbs.
Please add to this list:

1. implementation language: C/C++/others?
2. target architectures: posix/Win32/others?
3. threads: emulated in-core? using native threads? no threads at all?
4. relationship of 'soft code' to 'native code': VM? native bindings?
transparent interoperation? soft code->native compilation? multiple language
support?
5. module system: what goes in the core, and what comprises a module?
how do modules interact? 
6. distribution: should an explicit design goal be to support a distributed
mud? what are the requirements?
7. database: disk-based DB? distributed DB? is this invisible to the
programmer? what is the DB model (object vs relational vs ...)? could there be
multiple DB's?
8. memory management: garbage collection? explicit allocation/deallocation? 
how does this work with threading, DB, and the native/softcode divide? what
gc algorithm to use?
9. overall goals. 'a really fast mud'? 'a really scalable mud'? 'a really
usable mud'? 'a really advanced mud'?

clearly there are some interdependencies - choice of gc strategy is highly
sensitive to the presence or absence of threads, the db might be closely tied
to threading and gc, etc. I don't even know where to start with
network-distributed gc.

James